CAJ | 학술논문

由于SURF算子对光照仅有部分的不变性，从而在光照条件变化大的环境下可能会导致提取的特征数与匹配数减少的不足。为了克服该不足，文中深入分析光照变化对SURF算子对特征提取的影响，并提出利用对角线偏移模型对SURF算子进行改进，统一输入图像的光照信息，使得改进后的特征算子对光照的变化具有更强的不变性，从而使得改进后的算法在较大光照变化的条件下依旧能够稳定地提取特征点及匹配。
유우SURF산자대광조부유부분적불변성，종이재광조조건변화대적배경하가능회도치제취적특정수여필배수감소적불족。위료극복해불족，문중심입분석광조변화대SURF산자대특정제취적영향，병제출이용대각선편이모형대SURF산자진행개진，통일수입도상적광조신식，사득개진후적특정산자대광조적변화구유경강적불변성，종이사득개진후적산법재교대광조변화적조건하의구능구은정지제취특정점급필배。
Since SURF only is partly invariant for illumination,if it is applied into the variable illumination changing scene,the perform-ance of features and matching will decrease highly. In order to overcome the drawback,make a deeper analysis about how the illumination influences the performance of SURF,and improve the original SURF with uniform illumination by combining the diagonal-offset model. The improved algorithm has a stronger ability about invariant for illumination,so that can get stable feature points even under a big illumi-nation changing and match.